LensSwap

landing page
paste any news article link here and click 'submit'

Inspiration

Most of us don’t realize how often we are trapped inside the blind spots of our own biases and thinking. We skim articles, scroll timelines, and absorb narratives without noticing the missing context or the invisible assumptions underneath them, all the while thinking we are correct and anyone else, wrong or misinformed. I wanted to build something that helps people pause, zoom out, and see ideas from multiple angles -- not as a fact-checking tool (even they have , but as an individual critical-thinking companion that encourages epistemic humility and reduces unintentional echo-chamber thinking.

LensSwap was born out of that impulse: a tool that makes you smarter, fairer, and harder to manipulate.

What it does

paste article ==> get linked coverage from different POVs + a neutral, first-principles critique + gap-filling/counterpoint synthesis

LensSwap takes any article link or pasted text and instantly generates:

Multi-Perspective Coverage

Automatically finds how different outlets across the spectrum (left, center, right) cover the same topic — helping you see the full “idea space.”

First-Principles Argument Mapping

Extracts: core claims evidence logical fallacies confidence levels missing perspectives (“blind spots”)

Steelmanned Counterarguments

If opposing coverage exists, it uses it. If not, LensSwap generates the most charitable, rational counterpoints based on general principles and patterns in public discourse.

Common Ground Finder

Highlights facts that all sides actually agree on — something humans rarely notice.

Critical-Thinking Dashboard

A clean, tappable interface that acts like a personal reasoning coach: “Here’s what you missed. Here’s what questions to ask. Here’s how to evaluate it.”

This makes LensSwap useful not just for news, but for:

university assignments

debate prep

research

bias awareness

breaking personal echo chamber

How we built it

Frontend: React + Vite, handcrafted UI inspired by argument-mapping tools rather than news aggregators. Backend: Node + Express for stable, predictable topic extraction and article retrieval. LLM Analysis: Groq (Llama-3.3-70B) with carefully engineered prompts emphasizing: steelmanning claim/evidence separation stasis theory (fact/definition/policy) epistemic humility Content Extraction: Readability.js + JSDOM for robust article text parsing. Cross-Spectrum Retrieval: NewsAPI (with domain bucketing) to ensure balanced left/center/right sourcing. Bias Balancing: Hybrid model - blends outlet priors with LLM stance detection, preventing wild misclassifications.

Challenges we ran into

Designing prompts that encourage true first-principles reasoning instead of generic summaries. Getting balanced coverage: many APIs over-represent certain outlets; we had to build a custom left/center/right retrieval system. Removing paywalled or low-quality sources while keeping speed fast for hackathon use. Ensuring the LLM always produces JSON (painful). Making the UI feel intuitive instead of like a wall of text.

Accomplishments that we're proud of

Built a tool that genuinely improves critical thinking, not just content consumption. Achieved highly stable, clean JSON despite fast LLM calls. Created a UI that feels like a thinking assistant, not another newsreader. Engineered fallback logic so LensSwap always gives you something useful, even when no related articles exist.